Former Tesla AI Director Andrej Karpathy Joins Anthropic

Andrej Karpathy, a co-founder of OpenAI and the longtime leader of Teslas computer vision team, has joined Anthropic.

The announcement came directly from Karpathy on social media on Tuesday.

"Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D,"

He shared the update on X (link). He added that while he is jumping back into core research, he remains deeply passionate about education and plans to resume his work on it in time.

Accelerating Pre-Training Research

Karpathy is joining Anthropics pre-training team under team lead Nick Joseph. Pre-training is the compute-intensive phase during which a large language model such as Claude builds its foundational knowledge before fine-tuning.

According to Anthropic, Karpathy will assemble a new group focused on using Claude itself to speed up pre-training research. The move signals an emphasis on architectural optimizationrather than relying solely on additional computeto remain competitive with OpenAI and Google.

A Look Back at Karpathys Legacy

Karpathy served as Teslas Director of AI, leading the Autopilot and Full Self-Driving computer vision teams during a pivotal period. He announced his departure from Tesla in the summer of 2022 to take a break and pursue personal projects.

Since leaving the automaker, he has expressed interest in working at Tesla again, including a desire to contribute to the Optimus humanoid robot project. Elon Musk publicly invited him to return last summer.

Karpathy later returned to OpenAI for a year before departing in early 2024 to launch Eureka Labs, an AI-powered education startup. While his new position at Anthropic places Eureka Labs on hold, it also reflects how rapidly the artificial general intelligence landscape is shifting. He has often shared perspectives on the future of AGI, viewing it as an inevitable transition in human technology.

Anthropic Continues Its Talent Hunt

Karpathy isnt the only notable addition to Anthropic this week. The company also hired cybersecurity expert Chris Rohlf to join its frontier red team. Rohlf spent six years at Meta and previously worked on Yahoos well-known The Paranoids security team; he will be responsible for stress-testing Anthropics advanced models against severe digital threats.

By bringing in leading machine learning talent alongside seasoned security professionals, Anthropic is reinforcing its position at the frontier of AI development. How Karpathys background in physical computer vision at Tesla will influence text and multimodal systems like Claude remains to be seen, but his arrival makes Anthropic a key organization to watch in the months ahead.